Monitor Calico component metrics

Big picture

Use Prometheus configured for Calico components to get valuable metrics about the health of Calico.

Value

Using the open-source Prometheus monitoring and alerting toolkit, you can view time-series metrics from Calico components in the Prometheus or Grafana interfaces.

Features

This how-to guide uses the following Calico features:

Felix, Typha, and kube-controllers components configured with Prometheus configuration parameters (for consumption by Prometheus).

Concepts

About Prometheus

The Prometheus monitoring tool scrapes metrics from instrumented jobs and displays time series data in a visualizer (such as Grafana). For Calico, the “jobs” that Prometheus can harvest metrics from are the Felix and Typha components.

About Calico Felix, Typha, and kube-controllers components

Felix is a daemon that runs on every machine that implements network policy. Felix is the brains of Calico. Typha is an optional set of pods that extends Felix to scale traffic between Calico nodes and the datastore. The kube-controllers pod runs a set of controllers which are responsible for a variety of control plane functions, such as resource garbage collection and synchronization with the Kubernetes API.

You can configure Felix, Typha, and/or kube-controllers to provide metrics to Prometheus.

Before you begin…

In this tutorial we assume that you have completed all other introductory tutorials and possess a running Kubernetes cluster with Calico. You can either use kubectl or calicoctl to perform the following steps. Depending on which tool you would like to use, make sure you have the necessary prerequisites as shown below.

• kubectl
• calicoctl

If you wish to modify Calico configurations with kubectl binary you need to make sure you have the Calico API server in your cluster. The API server allows you to manage resources within the projectcalico.org/v3 api group.

note

Operator based installs include the API server by default.

For more information about the API server please use this link.

You can run calicoctl on any host with network access to the Calico datastore as either a binary or a container to manage Calico APIs in the projectcalico.org/v3 API group.

For more information about calicoctl please use this link.

How to

This tutorial will go through the necessary steps to implement basic monitoring of Calico with Prometheus.

1. Configure Calico to enable the metrics reporting.
2. Create the namespace and service account that Prometheus will need.
3. Deploy and configure Prometheus.
4. View the metrics in the Prometheus dashboard and create a simple graph.

1. Configure Calico to enable metrics reporting

Felix configuration

Felix prometheus metrics are disabled by default.

note

A comprehensive list of configuration values can be found at this link.

Use the following command to enable Felix metrics.

• kubectl
• calicoctl

kubectl patch felixconfiguration default --type merge --patch '{"spec":{"prometheusMetricsEnabled": true}}'

You should see an output like below:

felixconfiguration.projectcalico.org/default patched

calicoctl patch felixconfiguration default  --patch '{"spec":{"prometheusMetricsEnabled": true}}'

You should see an output like below:

Successfully patched 1 'FelixConfiguration' resource

Creating a service to expose Felix metrics

Prometheus uses Kubernetes services to dynamically discover endpoints. Here you will create a service named felix-metrics-svc which Prometheus will use to discover all the Felix metrics endpoints.

note

Felix by default uses port 9091 TCP to publish its metrics.

• Operator
• Manifest

kubectl apply -f - <<EOF
apiVersion: v1
kind: Service
metadata:
  name: felix-metrics-svc
  namespace: calico-system
spec:
  clusterIP: None
  selector:
    k8s-app: calico-node
  ports:
  - port: 9091
    targetPort: 9091
EOF

kubectl apply -f - <<EOF
apiVersion: v1
kind: Service
metadata:
  name: felix-metrics-svc
  namespace: kube-system
spec:
  clusterIP: None
  selector:
    k8s-app: calico-node
  ports:
  - port: 9091
    targetPort: 9091
EOF

Typha Configuration

• Operator
• Manifest

An Operator installation of Calico automatically deploys one or more Typha instances depending on the scale of your cluster. By default metrics for these instances are disabled.

Use the following command to instruct tigera-operator to enable Typha metrics.

kubectl patch installation default --type=merge -p '{"spec": {"typhaMetricsPort":9093}}'

You should see a result similar to:

installation.operator.tigera.io/default patched

note

Typha implementation is optional, if you don’t have Typha in your cluster you can skip Typha configuration section.

If you are uncertain whether you have Typha in your cluster execute the following code:

kubectl get pods -A | grep typha

If your result is similar to what is shown below you are using Typha in your cluster.

note

The name suffix of pods shown below was dynamically generated. Your typha instance might have a different suffix.

kube-system     calico-typha-56fccfcdc4-z27xj                         1/1     Running   0          28h
kube-system     calico-typha-horizontal-autoscaler-74f77cd87c-6hx27   1/1     Running   0          28h

You can enable Typha metrics to be consumed by Prometheus via two ways.

Creating a service to expose Typha metrics

note

Typha uses port 9091 TCP by default to publish its metrics. However, if Calico is installed using Amazon yaml file this port will be 9093 as its set manually via TYPHA_PROMETHEUSMETRICSPORT environment variable.

• Operator
• Manifest

kubectl apply -f - <<EOF
apiVersion: v1
kind: Service
metadata:
  name: typha-metrics-svc
  namespace: calico-system
spec:
  clusterIP: None
  selector:
    k8s-app: calico-typha
  ports:
  - port: 9093
    targetPort: 9093
EOF

kubectl apply -f - <<EOF
apiVersion: v1
kind: Service
metadata:
  name: typha-metrics-svc
  namespace: kube-system
spec:
  clusterIP: None
  selector:
    k8s-app: calico-typha
  ports:
  - port: 9093
    targetPort: 9093
EOF

kube-controllers configuration

Prometheus metrics are enabled by default on TCP port 9094 for calico-kube-controllers.

• Operator
• Manifest

The operator automatically creates a service that exposes these metrics.

You can use the following command to verify it.

kubectl get svc -n calico-system

You should see a result similar to:

calico-kube-controllers-metrics   ClusterIP   10.43.77.57     <none>        9094/TCP   39d

Creating a service to expose kube-controllers metrics

Create a service to expose calico-kube-controllers metrics to Prometheus.

kubectl apply -f - <<EOF
apiVersion: v1
kind: Service
metadata:
  name: kube-controllers-metrics-svc
  namespace: kube-system
spec:
  clusterIP: None
  selector:
    k8s-app: calico-kube-controllers
  ports:
  - port: 9094
    targetPort: 9094
EOF

Optionally, you can use the following command to modify the port by changing the KubeControllersConfiguration resource if desired.

note

Setting this value to zero will disable metrics in the kube-controllers pod.

• kubectl
• calicoctl

kubectl patch kubecontrollersconfiguration default --type=merge  --patch '{"spec":{"prometheusMetricsPort": 9095}}'

calicoctl patch kubecontrollersconfiguration default  --patch '{"spec":{"prometheusMetricsPort": 9095}}'

2. Cluster preparation

Namespace creation

Namespace isolates resources in your cluster. Here you will create a Namespace called calico-monitoring to hold your monitoring resources.

note

Kubernetes namespaces guide can be found at this link.

kubectl create -f -<<EOF
apiVersion: v1
kind: Namespace
metadata:
  name: calico-monitoring
  labels:
    app:  ns-calico-monitoring
    role: monitoring
EOF

Service account creation

You need to provide Prometheus a serviceAccount with required permissions to collect information from Calico.

note

A comprehensive guide to user roles and authentication can be found at this link.

kubectl apply -f - <<EOF
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: calico-prometheus-user
rules:
- apiGroups: [""]
  resources:
  - endpoints
  - services
  - pods
  verbs: ["get", "list", "watch"]
- nonResourceURLs: ["/metrics"]
  verbs: ["get"]
---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: calico-prometheus-user
  namespace: calico-monitoring
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: calico-prometheus-user
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: calico-prometheus-user
subjects:
- kind: ServiceAccount
  name: calico-prometheus-user
  namespace: calico-monitoring
EOF

3. Install prometheus

Create prometheus config file

We can configure Prometheus using a ConfigMap to persistently store the desired settings.

note

A comprehensive guide about configuration file can be found at this link.

• Operator
• manifest

kubectl apply -f - <<EOF
apiVersion: v1
kind: ConfigMap
metadata:
  name: prometheus-config
  namespace: calico-monitoring
data:
  prometheus.yml: |-
    global:
      scrape_interval:   15s
      external_labels:
        monitor: 'tutorial-monitor'
    scrape_configs:
    - job_name: 'prometheus'
      scrape_interval: 5s
      static_configs:
      - targets: ['localhost:9090']
    - job_name: 'felix_metrics'
      scrape_interval: 5s
      scheme: http
      kubernetes_sd_configs:
      - role: endpoints
      relabel_configs:
      - source_labels: [__meta_kubernetes_service_name]
        regex: felix-metrics-svc
        replacement: $1
        action: keep
    - job_name: 'typha_metrics'
      scrape_interval: 5s
      scheme: http
      kubernetes_sd_configs:
      - role: endpoints
      relabel_configs:
      - source_labels: [__meta_kubernetes_service_name]
        regex: typha-metrics-svc
        replacement: $1
        action: keep
      - source_labels: [__meta_kubernetes_pod_container_port_name]
        regex: calico-typha
        action: drop
    - job_name: 'kube_controllers_metrics'
      scrape_interval: 5s
      scheme: http
      kubernetes_sd_configs:
      - role: endpoints
      relabel_configs:
      - source_labels: [__meta_kubernetes_service_name]
        regex: calico-kube-controllers-metrics
        replacement: $1
        action: keep
EOF

kubectl apply -f - <<EOF
apiVersion: v1
kind: ConfigMap
metadata:
  name: prometheus-config
  namespace: calico-monitoring
data:
  prometheus.yml: |-
    global:
      scrape_interval:   15s
      external_labels:
        monitor: 'tutorial-monitor'
    scrape_configs:
    - job_name: 'prometheus'
      scrape_interval: 5s
      static_configs:
      - targets: ['localhost:9090']
    - job_name: 'felix_metrics'
      scrape_interval: 5s
      scheme: http
      kubernetes_sd_configs:
      - role: endpoints
      relabel_configs:
      - source_labels: [__meta_kubernetes_service_name]
        regex: felix-metrics-svc
        replacement: $1
        action: keep
    - job_name: 'typha_metrics'
      scrape_interval: 5s
      scheme: http
      kubernetes_sd_configs:
      - role: endpoints
      relabel_configs:
      - source_labels: [__meta_kubernetes_service_name]
        regex: typha-metrics-svc
        replacement: $1
        action: keep
    - job_name: 'kube_controllers_metrics'
      scrape_interval: 5s
      scheme: http
      kubernetes_sd_configs:
      - role: endpoints
      relabel_configs:
      - source_labels: [__meta_kubernetes_service_name]
        regex: kube-controllers-metrics-svc
        replacement: $1
        action: keep
EOF

Create Prometheus pod

Now that you have a serviceaccount with permissions to gather metrics and have a valid config file for your Prometheus, it’s time to create the Prometheus pod.

kubectl apply -f - <<EOF
apiVersion: v1
kind: Pod
metadata:
  name: prometheus-pod
  namespace: calico-monitoring
  labels:
    app: prometheus-pod
    role: monitoring
spec:
  serviceAccountName: calico-prometheus-user
  containers:
  - name: prometheus-pod
    image: prom/prometheus
    resources:
      limits:
        memory: "128Mi"
        cpu: "500m"
    volumeMounts:
    - name: config-volume
      mountPath: /etc/prometheus/prometheus.yml
      subPath: prometheus.yml
    ports:
    - containerPort: 9090
  volumes:
  - name: config-volume
    configMap:
      name: prometheus-config
EOF

Check your cluster pods to assure pod creation was successful and prometheus pod is Running.

kubectl get pods prometheus-pod -n calico-monitoring

It should return something like the following.

NAME             READY   STATUS    RESTARTS   AGE
prometheus-pod   1/1     Running   0          16s

4. View metrics

You can access prometheus dashboard by using port-forwarding feature.

kubectl port-forward pod/prometheus-pod 9090:9090 -n calico-monitoring

Browse to http://localhost:9090 you should be able to see prometheus dashboard. Type felix_active_local_endpoints in the Expression input textbox then hit the execute button. Console table should be populated with all your nodes and quantity of endpoints in each of them.

note

A list of Felix metrics can be found at this link. Similar lists can be found for kube-controllers and Typha.

Push the Add Graph button, You should be able to see the metric plotted on a Graph.

Cleanup

This section will help you remove resources that you have created by following this tutorial. Please skip this step if you like to deploy Grafana to Visualize component metrics. First remove the services by executing the following command:

• Operator
• Manifest

kubectl delete service felix-metrics-svc -n calico-system
kubectl delete service typha-metrics-svc -n calico-system

kubectl delete service felix-metrics-svc -n kube-system
kubectl delete service typha-metrics-svc -n kube-system
kubectl delete service kube-controllers-metrics-svc -n kube-system

Return Calico configurations to their default state.

• kubectl
• calicoctl

kubectl patch felixConfiguration default --type merge --patch '{"spec":{"prometheusMetricsEnabled": false}}'
kubectl patch installation default --type=json -p '[{"op": "remove", "path":"/spec/typhaMetricsPort"}]'

calicoctl patch felixConfiguration default  --patch '{"spec":{"prometheusMetricsEnabled": false}}'

Finally, remove the namespace and RBAC permissions.

kubectl delete namespace calico-monitoring
kubectl delete ClusterRole calico-prometheus-user
kubectl delete clusterrolebinding calico-prometheus-user

Best practices

If you enable Calico metrics to Prometheus, a best practice is to use network policy to limit access to the Calico metrics endpoints. For details, see Secure Calico Prometheus endpoints.

If you are not using Prometheus metrics, we recommend disabling the Prometheus ports entirely for more security.

Next Steps

Visualizing metrics via Grafana.